CN110191248B

CN110191248B - Feedback-based unmanned aerial vehicle image transmission method of Bats Code

Info

Publication number: CN110191248B
Application number: CN201910524818.6A
Authority: CN
Inventors: 王臣玺; 史治平; 孙红霞; 张茜
Original assignee: Public Security Fire Forces College; Tianfu New District Southwest Jiaotong University Research Institute
Current assignee: Public Security Fire Forces College; Tianfu New District Southwest Jiaotong University Research Institute
Priority date: 2019-06-07
Filing date: 2019-06-07
Publication date: 2020-09-29
Anticipated expiration: 2039-06-07
Also published as: CN110191248A

Abstract

The invention belongs to the technical field of unmanned aerial vehicle communication, and relates to an unmanned aerial vehicle image transmission method based on feedback Bats Code. The invention relates to an application design of low-complexity network codes (Bats codes) based on feedback in unmanned aerial vehicle image transmission, which is suitable for a network layer and an upper layer thereof. The Bats Code has high reliability and low complexity, but has the defects of large coefficient overhead, large decoding overhead in the later decoding period and the like, so that various problems exist in practical application. Therefore, the invention provides an unmanned aerial vehicle image transmission scheme based on feedback Bats Code aiming at the two main defects. Compared with direct transmission, the scheme improves the reliability of transmission; compared with the traditional Bats Code, the decoding success rate can be further improved through a certain number of feedback, the coefficient overhead is reduced, and the transmission efficiency is improved.

Description

Feedback-based unmanned aerial vehicle image transmission method of Bats Code

Technical Field

The invention belongs to the technical field of unmanned aerial vehicle communication, and relates to an unmanned aerial vehicle image transmission method based on feedback Bats Code.

Background

Under the influence of increasingly complex electromagnetic environments, image transmission is taken as a key part of an unmanned aerial vehicle communication system, and the performance of the unmanned aerial vehicle is directly influenced by the anti-interference performance of the image transmission. At present, the communication distance of most unmanned aerial vehicles is 5-10 km, and the limitation of the communication distance leads to that the unmanned aerial vehicles or the unmanned aerial vehicles and a ground receiver cannot directly carry out remote communication, so that the communication is realized through a relay. Thus, long-range drone communication systems are generally made up of a drone, a relay (drone or other device) and a ground receiver or drone. Considering the characteristics of simple channel coding algorithm, low complexity, high reliability, high real-time performance and the like required by unmanned aerial vehicle communication, a feasible unmanned aerial vehicle image transmission method is designed by researching and analyzing a channel coding scheme suitable for remote unmanned aerial vehicle communication, and the method has very important functions and significance for improving the reliability of signals transmitted by an unmanned aerial vehicle in a wireless channel and improving the performance of an unmanned aerial vehicle communication system.

At present, Turbo codes, LDPC codes and the like are often used for channel coding between an unmanned aerial vehicle and a ground receiver. However, Turbo codes and LDPC codes have the problems of high complexity, poor real-time performance, high hardware cost, and the like. For long-distance communication, the direct storage and forwarding mode adopted at the relay node can cause the accumulation of data loss, thereby causing the decoding performance to be poor. Therefore, a joint channel coding scheme based on the LDPC code and the physical layer network coding is proposed, so as to solve the problem of data loss accumulation at the relay node. However, the scheme has the problems of high complexity, high coefficient overhead and error floor of the LDPC code under high signal-to-noise ratio. The bulk sparse Code (Bats Code) serving as a joint coding scheme based on the fountain Code and the network coding can solve the problems, and has the advantages of no Code rate, low complexity, high throughput, high safety and the like, but the application of the bulk sparse Code to unmanned aerial vehicle communication still faces many problems. The most important problem is that the coefficient overhead is large, resulting in low transmission efficiency.

Disclosure of Invention

The invention aims to provide an unmanned aerial vehicle image transmission scheme based on feedback Bats Code, which is suitable for a network layer and an upper layer thereof.

For the sake of understanding, the following description will first be made on a part of the technical principle employed in the present invention.

The Bats Code is a new coding and decoding scheme combining fountain Code and network coding, and consists of inner Code and outer Code. Wherein, the outer code is fountain code coding, and the inner code is random linear network coding. The basic idea is that at the sending end, the information data to be transmitted is divided into source packets with specific length, the source packets are selected according to the degree distribution function and fountain code coding is carried out on the source packets, and the data packets with batch as the unit are generated and transmitted on the channel. And at the intermediate node, carrying out random linear network coding inside the batch on the received batch, and then transmitting the batch to the receiving end. After the transmission through the channel, the receiving end can successfully decode the source information to be transmitted as long as the receiving end can receive a sufficient number of batchs without considering the channel condition. The basic codec format of the Bats Code will be mainly described below.

Bats Code coding

The Bats Code encoding is divided into outer Code encoding and inner Code encoding. The outer code coding is fountain code coding and is carried out at an information source node; the inner code coding is random linear network coding and is carried out at the intermediate node. The whole encoding process of the Bats Code is shown in fig. 1.

The process of generating a batch by outer code encoding can be expressed as:

X_i＝B_iG_i

wherein B is_iFor d packets, G, randomly selected according to the degree distribution_iTo generate the matrix, the dimensions are d × M, X_iIs the generated batch.

The inner code coding adopts random network coding, namely, the intermediate node carries out random linear combination on the received coding packets belonging to the same batch again, thereby generating a new data packet. The encoding process is represented as:

Y_i＝X_iH_i＝B_iG_iH_i

in the above formula, H_iIs a randomly generated transfer matrix with dimension M × M, M being middleThe number of the coding packets received by the intermediate node; y is_iA new batch generated after re-encoding.

Bats Code decoding

The Bats Code decoding employs an iterative algorithm. At each step of decoding, the decoder looks for, among the received bats, those having a degree equal to the rank of its coefficient matrix, and the set of these bats is called the output interpretable set. The set of packets they join is called the input translatable set. After that, the decoder makes linear combination of each decoded information packet and all the code packets connected with it, and the calculation result replaces the original value of the corresponding code packet, and after that, the connection relation between them is deleted. The above process is repeated until there is no batch with a degree equal to the rank of its coefficient matrix. If all data packets are recovered, the decoding is successful, otherwise, the decoding fails.

The BP decoding algorithm is essentially a low complexity matrix inversion method. Suppose that the transmitted data packet and the received coded packet vector are X ═ X respectively₁,x₂,...,x_k]^TAnd Y ═ Y₁,y₂,...,y_n]^T. The receiving end can reconstruct the coefficient coding matrix G × H of the Bats Code according to a certain protocol (e.g., header information). Thus is provided with

Y＝G·H·X

If can calculate (GH)^-1Then can be determined by X ═ GH^-1Y calculates the original data packet vector X, which can obtain the effect of maximum likelihood decoding and decode the most data.

The invention provides an unmanned aerial vehicle image transmission scheme based on feedback Bats Code, which aims to overcome the defects of high complexity and high coefficient overhead of traditional network coding and improve the decoding success rate of a receiving end under the condition of limited feedback. The reduction of the coefficient overhead is mainly realized by fixing the initial states of the pseudo-random sequence generators at three points of a sending end, a relay and a receiving end and avoiding the transmission of parameters such as coefficient vectors, associated packet numbers and the like. The main idea of feedback is: the sending end sends a certain amount of Bats Code encoding packets according to the channel condition, and the receiving end starts decoding after receiving is finished. If the decoding is not successful, the receiving end selects the packet with the maximum degree from the undeciphered information packets and feeds back the packet number to the transmitting end. When the sending end receives the fed back packet number but does not receive the ACK, the sending end continues to send the Bats Code encoding packet and directly transmits the original data packet corresponding to the feedback to the receiving end in the form of batch with the degree of 1. If the decoding is successful, the receiving end feeds back an ACK signal, and the transmission is finished.

The specific steps of the design scheme are as follows:

1. a sending end:

a. and segmenting and packaging the acquired images. Suppose that the RGB three-dimensional data of the image to be transmitted is D_1(N,T,L)The original information is equally divided into K information packets, each data packet contains L symbols, and the symbol value is in finite field 2^qWithin the range. Where K is referred to as the code length of the coding scheme.

b. And judging whether feedback information exists or not. If the feedback information is an ACK signal, stopping coding and sending; if no feedback information or the feedback information is a packet number (1-K), fountain code coding (outer code coding) is carried out on the packet according to the following steps to generate N_j(the jth encoding generates the number of batchs) and transmits. For the case where the feedback information is a packet number, except for transmission of N_jBesides 1 batch, the information packet corresponding to the feedback packet number is directly transmitted in the form of batch with the degree of 1 without participating in the Bats Code coding. The above steps are shown in fig. 2.

(1) An integer d is randomly selected within the range of 1-K according to a certain distribution omega (called encoding degree distribution). Coding degree distribution omega ═ omega₁,Ω₁,...,Ω_D]The probability of the random selected degree i is shown as omega_i. Where D is the maximum degree, which generally does not exceed the code length K, the degree distribution can also be expressed in the form of a function, i.e.:

Ω(x)＝Ω₁x+Ω₂x²+...+Ω_dx^D

this function is called degree distribution function, where the index of x represents the degree of choice, Ω_iThe probability is represented.

(2) D different packets are uniformly and randomly selected from the K information packets;

(3) in a limited domain 2^qAnd randomly generating a (d, M) coefficient matrix G (generating matrix) in the range, and linearly combining the symbols in the d packets according to column information in the matrix G to generate M coding packets to form a batch. Where M is the length of the batch, and d is called the degree of the batch.

The outer code coded packet structure is shown in fig. 3. The outer Code encoding Packet consists of a Batch number (Batch ID), a Packet number (Packet ID), a Degree (granularity), a Code word (Code), and a Parity bit (Parity digit).

Batch ID: the number of the batch designates the serial number of the batch to which the code packet belongs.

Packet ID: the packet number of the packet is encoded. For the code packet, the position of the code packet in the batch is marked, namely 1-M; for the direct transmission packet, the corresponding is the packet number, i.e. 1 to K.

Degree: the degree of the encoded packet indicates the degree of the batch in which the encoded packet is located.

Part digit: and the odd check bits are used for ensuring the correctness of the coded packet in the transmission process.

The random packet selection and the coefficient matrix generation in the steps (2) and (3) are realized by a pseudo-random sequence generator.

2. Relaying:

a. and carrying out parity check on the received coded packet, and discarding the coded packet if the parity check is wrong.

b. The received encoded packet is divided into packets in units of batch according to the batch ID.

c. In each batch, counting the coding packets lost from the source to the relay stage according to the Packet ID, and recording the Packet loss condition in all the coding packets in the batch by using a Flag symbol. The statistical and recording process is as follows: and (3) recording the packet loss condition of the coded packets in the batch by using an array Flag _ loss1 with the length of M, wherein if the ith coded packet is lost, the Flag _ loss1(i) is equal to 0, and otherwise, the Flag _ loss1(i) is equal to 1. The 8-bit binary number is converted into decimal number, which is marked as Flag symbol and stored in an inner code encoding packet. For the encoding batch, the operation is as above; for the direct transfer batch, the Flag symbol is set to 0 directly.

d. The code packets belonging to the same batch are subjected to random linear network coding (or inner code coding), as shown in fig. 4. And (3) randomly generating a (k, M) transfer matrix H by a pseudo-random sequence generator to linearly combine the coded packets in the batch to generate M coded packets. Wherein k is the number of correctly received code packets in the batch. The parity bits are updated after encoding. For the pass-through batch, no random linear network coding and no parity bit update is performed.

The inner code coded packet structure is shown in fig. 5. The inner code encoding packet adds a symbol, namely Flag symbol, on the basis of the outer code encoding packet, so that the packet loss condition from the information source to the relay stage can be transmitted to the information sink.

3. Signal sink

b. The received encoded packet is divided into packets in units of batch according to the batch ID, and the direct transmission batch is separated from the encoded batch by recognizing the Flag symbol.

c. For encoding batch, the process of passing its coefficient matrix through the channel is simulated.

(1) And recovering the packet loss condition from the information source to the relay stage in the batch according to the symbol Flag symbol of the encoding packet header, namely converting the Flag symbol from decimal to M-bit binary number, and storing the binary number in an array Flag _ loss 1.

(2) And counting Packet loss conditions in the stage from the relay to the sink in the batch according to the Packet ID, and recording by using an array Flag _ loss2 with the length of M, wherein if the ith coded Packet is lost, the Flag _ loss2(i) is 0, and otherwise, the Flag _ loss2(i) is 1.

(3) For each batch, setting the initial state of the pseudo-random sequence generator as the initial state of the pseudo-random sequence generator at the information source and the initial state of the pseudo-random sequence generator at the relay respectively, thereby obtaining a generating matrix G and an associated packet number Pos which are the same as those of the information source and a transfer matrix H which is the same as those of the relay. Wherein Pos refers to the packet number of the original data packet participating in the batch encoding during the outer code encoding, and is used for BP decoding.

(4) The channel is simulated. In the erasure channel, if a coded packet is lost, all the information carried by the coded packet, including the coefficient vector, will also be lost. Therefore, the process of passing the jth batch coefficient matrix through the channel can be considered as follows: g_j·e_1j·H_j·e_2j. Wherein e is_1j、e_2jThe unit arrays are diagonal elements Flag _ loss1 and Flag _ loss2, respectively.

d. The receiving end receives N_jAfter batch, BP decoding is started. The specific decoding steps (fig. 6) are as follows:

1. judging whether a direct transmission data packet is received or not, if so, regarding the packet as a translated data packet, and performing the step 3; if not, the rank and degree of the search coefficient matrix are the same (rank (G)_j·e_1j·H_j·e_2j)＝d_j) The batch of (1); if a batch meeting the condition is found, the step 2 is carried out; otherwise, decoding is finished;

2. solving a linear equation set by a Gaussian elimination method for decoding;

3. based on Pos, the batch associated with the translated packet is updated. If the decoded packet i is related to the batch j, sequentially performing the following three steps to update the batch j:

1) subtracting 1 from the degree of batch j;

2) eliminating the correlation between the undeciphered packet in the batch j and the pabkey i;

3) matrix G of coefficients of batch j_j·e_1j·H_j·e_2jAnd eliminating the row corresponding to the middle packet i.

4. And after the updating is finished, repeating the step 1 until the decoding is finished.

e. And (6) feeding back. If the decoding is successful, feeding back an ACK signal; if the information packet fails, one packet with the maximum degree is searched in the undeciphered information packets, and the packet number of the packet is fed back to the sending end.

The invention has the advantages that compared with direct transmission, the reliability of transmission is improved; compared with the traditional BatsCod, the pseudo-random sequence is used, so that the transmission of parameters such as coefficient vectors and the like is avoided, the coefficient overhead is reduced, the transmission efficiency is improved, and the decoding success rate is further improved through a certain number of feedbacks.

Drawings

FIG. 1 is a schematic diagram of Bats Code encoding;

fig. 2, fig. 4, and fig. 6 are data processing flowcharts of a transmitting end, a relay, and a receiving end, respectively, and correspond to outer code encoding, inner code encoding, and BP decoding and feedback, respectively;

fig. 3, fig. 5 and fig. 7 are schematic structural diagrams of an outer code encoded packet, an inner code encoded packet and a direct transmission packet, respectively;

fig. 8 is a graph of redundancy versus decoding success rate.

Detailed Description

The present invention will be further described with reference to the following examples

Examples

The embodiment is an unmanned aerial vehicle image transmission method based on feedback Bats Code, wherein the Bats Code is a coding and decoding scheme combining fountain Code and network Code, and the image transmission method comprises the following steps:

a. convert the image of unmanned aerial vehicle A's camera collection into two-dimensional data by three-dimensional data and carry out partial shipment group dress, D promptly_(L,T,N)→D′_(L,K)Where K ═ T × N denotes the number of columns of the two-dimensional data; n represents the number of three primary colors, equal to 3; l, T respectively indicating the number of rows and columns of the sub-matrix corresponding to each color; packaging the obtained two-dimensional data into packets according to columns so as to obtain K original information packets with the length of L; calculating corresponding degree distribution according to the estimated channel condition, the number K of the information packets and the length M of the batch; and then carrying out 1 st encoding transmission, generating a generating matrix G and packet numbers of the packets participating in encoding by using a pseudo-random sequence generator, carrying out fountain code encoding on the selected packets, and forming a batch by taking M encoded packets as a unit, wherein the generation process of the ith batch is represented as follows:

X_i＝B_i·G_i

wherein B is_iFor d packets, G, randomly selected according to degree distribution_iTo generateMatrix with dimensions d × M, X_iIs the generated batch;

the unmanned aerial vehicle A broadcasts a certain number of coding packets according to the coding method and the channel estimation condition;

b. after carrying out parity check on the received coded packets, the relay unmanned aerial vehicle B carries out grouping by taking batch as a unit; after grouping, firstly counting packet loss information from a signal source to a relay stage in batch, and recording as Flag symbol; then, a transfer matrix H is generated by using a pseudo-random sequence generator, random linear network coding is realized by linear combination of the coding packets in the batch, and M coding packets are regenerated; the inner code encoding process of the ith batch is expressed as follows:

Y_i＝X′_i·H_i＝(B_i·G_i)·E_loss1·H_i

wherein X'_iFor relaying the received ith batch, E_loss1Is a unit matrix with the diagonal element Flag symbol, H_iIs a transfer matrix with dimension dim (X'_i,2)×M，Y_iIs the regenerated batch;

c. after receiving the coded packet and performing parity check, the ground receiver or the unmanned aerial vehicle C performs grouping by taking batch as a unit; then, counting packet loss information of a relay to receiver stage in the batch, and recording as Flag loss; then, setting the initial state of the pseudo-random sequence generator as the initial state of the pseudo-random sequence generator at the information source and the initial state of the pseudo-random sequence generator at the relay respectively, and generating a generating matrix G and associated packet information Pos which are the same as those at the information source and a transfer matrix H which is the same as those at the relay; simulating the process of passing the channel by the coefficient matrix according to the Flag symbol and the Flag loss; coefficient matrix GH of ith batch_iIs shown below

GH_i＝G_i·E_loss1·H_i·E_loss2

Wherein E is_loss2A unit array with a diagonal element Flag loss;

after receiving the transmitted coded packet, the ground receiver or the receiving unmanned aerial vehicle starts BP decoding, namely searches for a batch with the rank equal to degree, decodes the BP by Gaussian elimination after finding, and then packs the decoded information into other associated batches for elimination and updating; after updating, executing the step c again until the batch meeting the rank equality degree cannot be found, and ending the decoding;

after the decoding is finished, converting the decoded data of the information packet from two-dimensional data into three-dimensional data, and recovering the image; if the image is completely recovered, feeding back an ACK signal to the unmanned aerial vehicle A; if the image is not completely recovered, the ground receiver or the receiving unmanned aerial vehicle searches for the information packet with the maximum degree in the unrevealed information packet, and feeds back the packet number to the unmanned aerial vehicle A;

d. if the unmanned aerial vehicle A receives the feedback ACK signal, stopping coding and sending a coded packet; if the received packet number is the packet number, the 2 nd encoding transmission is carried out: broadcasting the information packet together with a certain number of new coding packets in a batch mode with the degree of 1; for the ith transmitted coded packet, i is more than or equal to 2, and the processing process of the relay receiver B on the ith transmitted coded packet is consistent with that of the step B; after parity check is carried out on the direct transmission batch, the direct transmission is carried out without inner code coding; the receiving process of the ground receiver or the unmanned aerial vehicle C is consistent with the operation of the step C, but when decoding, firstly, the received direct transmission data packet is regarded as a decoded information packet, together with the decoded information packet, the received batch is firstly subjected to primary elimination and updating, and then, BP decoding is carried out; c, after decoding, feeding back according to the principle in the step c; and d, circularly performing the steps a-d until the unmanned aerial vehicle A receives the ACK signal or the feedback times reach the set upper limit value.

The effect of the solution of the present invention is described below with reference to fig. 8.

It is assumed that the information length K is 10000, the packet length L is 60, the batch length M is 8, the packet loss rate e of the erasure channel is 0.2, and the range is 256.

First, the degree distribution Ψ is calculated from the information length K, the channel state e, and the batch length M. The degree distribution calculated according to K10000, M8, e 0.2 is as follows:

TABLE 1 degree distribution

i	8	11	14	16	23	27	38	41
									Ψ_i	0.3698	0.1636	0.035	0.1326	0.0711	0.0504	0.0295	0.0394
i	62	64	103	104	178	358	359
									Ψ_i	0.0138	0.0305	0.0137	0.0152	0.0192	0.0087	0.0074

According to the degree distribution and the known parameters, a direct transmission data packet (as shown in fig. 7) carrying 501 bits of data is obtained through calculation, and the effective data is 480 bits; the Bats Code encoding packets (as shown in FIGS. 3 and 5) carry 520 bits of data on average, and the effective data is 480 bits.

Second, the initial state of the pseudo-random sequence generators at the source and relay is determined. For each batch, the pseudo-random sequence generators at the source and the relay each correspond to an initial state. For example, if the total number N of transmitted batchs is 2000, the source and the relay store 2000 random numbers as the initial state of the pseudo random sequence generator, and the sink needs to store 4000 random numbers for decoding.

Then, the number of first sent batchs N1 is estimated according to the channel condition (the decoding success rate cannot be made too high or too low) so that the feedback has a more obvious forward effect. The transmitting end encodes and transmits N1 batchs, and the receiving end decodes and feeds back the maximum packet number in the undeciphered information packet after receiving. After the sending end receives the feedback, the sending end directly transmits the information packet corresponding to the feedback packet number and sends N_i(i is more than or equal to 2) code batchs. And the receiving end carries out decoding and feedback after receiving. And the above steps are repeated until the feedback times reach the upper limit or the decoding is successful.

In the simulation design, the redundancy r is 0.62, N1 is 1876 as the number of batchs transmitted for the first time, and the upper limit value of the number of feedbacks is 1, that is, feedbacks are performed only once. The effect of the feedback on the decoding is analyzed by increasing the redundancy, i.e. the number of taps N2 sent at 2 nd time. Under the condition of the parameters, 20 frames of images are simulated, and the average success rate is obtained to obtain the curve relation shown in fig. 8. As can be seen from FIG. 8, when the redundancy is low (r ≦ 0.67), the decoding success rate of the conventional Bats Code is not as good as that of the direct transmission. However, when the redundancy is higher than 0.67, the decoding success rate of the traditional Bats Code is better than that of the direct transmission, and the difference between the two is increased along with the increase of the redundancy, and the maximum can reach 15%. Therefore, under the condition of low Code rate, the performance of unmanned aerial vehicle image transmission by using the Bats Code is better than that of direct transmission. Subsequently, the decoding performance of the Bats Code with feedback and the traditional Bats Code are compared. As can be seen from FIG. 8, the decoding performance of the Bats Code with feedback is better than that of the traditional Bats Code when the redundancy is 0.62-0.72, and the difference and the redundancy between the two almost show a convex function relationship, wherein the peak value is reached between 0.66-0.67, and the difference is 12%. Because only one time is fed back and only one packet number is fed back, the time delay caused by the feedback is almost negligible. Therefore, under the condition of low Code rate, the feedback-based Bats Code is used as a channel coding scheme in the image transmission of the unmanned aerial vehicle, so that the complexity is low, the reliability is high, the safety is high, the decoding success rate can be further improved through a certain number of feedback, the coefficient overhead is reduced, and the transmission efficiency is improved.

Claims

1. An unmanned aerial vehicle image transmission method based on feedback Bats Code, wherein the Bats Code is a coding and decoding scheme combining fountain Code and network Code, the image transmission method comprises the following steps:

a. convert the image of unmanned aerial vehicle A's camera collection into two-dimensional data by three-dimensional data and carry out partial shipment group dress, D promptly_(L,T,N)→D′_(L,K)Where K ═ T × N denotes the number of columns of the two-dimensional data; n represents the number of three primary colors, equal to 3; l, T respectively indicating the number of rows and columns of the sub-matrix corresponding to each color; packing the obtained two-dimensional data into packets according to columns to obtain KOriginal information packets with length L; calculating corresponding degree distribution according to the estimated channel condition, the number K of the information packets and the length M of the batch; and then carrying out 1 st encoding transmission, generating a generating matrix G and packet numbers of the packets participating in encoding by using a pseudo-random sequence generator, carrying out fountain code encoding on the selected packets, and forming a batch by taking M encoded packets as a unit, wherein the generation process of the ith batch is represented as follows:

X_i＝B_i·G_i

wherein B is_iFor d packets, G, randomly selected according to degree distribution_iTo generate the matrix, the dimensions are d × M, X_iIs the generated batch;

Y_i＝X′_i·H_i＝(B_i·G_i)·E_loss1·H_i

c. after receiving the coded packet and performing parity check, the ground receiver or the unmanned aerial vehicle C performs grouping by taking batch as a unit; then, counting packet loss information of a relay to receiver stage in the batch, and recording as Flag loss; then, the initial state of the pseudo-random sequence generator is set to the initial state of the pseudo-random sequence generator at the source and the relay, respectively, and the same generation matrix G as that at the source is generatedThe associated package information Pos and the same transfer matrix H as the relay; simulating the process of passing through the channel by the coefficient matrix according to the Flag symbol and the Flag; coefficient matrix GH of ith batch_iIs shown below

GH_i＝G_i·E_loss1·H_i·E_loss2

Wherein E is_loss2A unit array with a diagonal element Flag loss;